Improving the accuracy of photovoltaic generation forecasting by considering particulate matter variables

Abstract

Particulate matter contributes to changing the environment and thus reduces solar irradiance. The reduction in solar irradiance is expected to reduce the power output of photovoltaic (PV) generation. Therefore, analyzing the influence of particulate matter on PV generation is required to understand and efficiently utilize PV systems. Furthermore, it is important to consider these influences when developing forecasting algorithms to improve the algorithm's accuracy. We first analyzed the influence of particulate matter on PV output by examining the correlation of particulate matter concentration on PV power output based on the Pearson correlation method. Then, the PV forecasting model including particulate matter variables and interaction variables was developed based on an artificial neural network (ANN). The results reveal that the particulate matter variables can help improve the accuracy of PV forecasting.